Moldflow Insight Forum
Welcome to Autodesk’s Moldflow Insight Forums. Share your knowledge, ask questions, and explore popular Moldflow Insight topics.
cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 

TIME TO COMPLETE COOLING ANALYSIS

28 REPLIES 28
SOLVED
Reply
Message 1 of 29
rohitr
3085 Views, 28 Replies

TIME TO COMPLETE COOLING ANALYSIS

Hello all,

 

I have a problem regarding time to complete cooling analysis.I am analysing simple component with about 65000 elements.

I have analyse the component using fil+flow+warp sequence.To complete the same analysis it hardly takes 4 to 5 hours.

 

now wish to analyse the component using cool+fill+flow+warp. but cooling anaysis take longer time to complete near about 18 hours or more.

The cooling circuit is simple without any defects.

Kindly let me know suggetions

28 REPLIES 28
Message 2 of 29
yannick.moret
in reply to: rohitr

Use a machine with enough RAM.

Cool need to write some matrix to solve the thermal solution.

If there is enough space in RAM, it will put it all in RAM, and analysis time will be fast.

If you don't have enoug RAM, it will put it on your hard disk, and then the time will be long as the CPU will have to get the data from it.

It is easy to check, if you look in task manager and see cool is not using 1 core full time, that mean it is runningon disk and will be pretty slow.

 

Look at start of screen output in cool :

Total Mbytes required for Cool analysis = 3628.13

Total Mbytes available for Cool analysis = 420143.38

Now beginning the task: Input part model

Current time is: Thu Aug 11 15:41:52 2011

 

So for example here I need 3.6 gig of memory.

 

Think at closing synergy to save some RAM also while your analysis is running, if you are a bit short on your machine.

Message 3 of 29
abakharev
in reply to: rohitr

Is it steady-state cool or transient?

Message 4 of 29
tobi_cologne
in reply to: abakharev

Hello!

 

I have no solution, but the same problem. I'm using a 12 gb machine so I think it has to be enough. I even have the problem with just filling analysis, it takes really much time (10 hours for dual domain model). When the analysis has completed, the message window "Analysis complete" appaers 10 or nearly 15 times, which I have to click "OK". I just installed AMI 2012 on my machine (OS: VIsta).

 

Okay thanks for your replies! 🙂

Message 5 of 29
raalteh
in reply to: rohitr

When you part has a lot of details and small elements, the Cool solver with boundary element method may get slowed down. If you are using a 3D mesh, and you are using Moldflow Insight 2012, try the new Transient Cooling capabilities (even while using it for a steady state analysis). This is likely going to converge much quicker. 

Hanno van Raalte,

Product Manager - Injection Molding & Moldflow products
Message 6 of 29
piotr.wozniacki
in reply to: raalteh

I have a very fast computer with 16GB RAM, and yet the second phase of Cool (FEM) 3D anaysis does take enormous emount of time to complete.

 

Even though the entire sequence is parallelized, and I check the option for using max number of threads - and all 8 threads are reported to be in use - only  the first phase of the Cool (FEM) 3D snalysis can tax my fasy i7 extreme CPU up to 100% for all 8 threads. The second phase - transient within cycle - only uses some 15-18% - and this is not due to low memory and swap file trashing!

 

Just why is that, and what's the bottleneck?

 

Piotr

Message 7 of 29
rohitr
in reply to: abakharev

hello all,

How to define steady state or transient cooling ?

 

Kindly let me know if any suggetions.

 

ROHIT

Message 8 of 29
HugoHerrera7042
in reply to: rohitr

Hi,

happy to find this discussion since i was going to start a very similar one.

 

I have a midplane model with cool+fill+pack+warp (38,000trias elements) PP 32% TF, and when I run the analysis it gets stucked at ~54% of cooling analysis...for days...withouth any error or warning message...now why is that happening??

 

i know in my experience that usually cooling analyses involving fiber filled materials take significantly longer, but i don't think i'm going anywhere with this iteration....anyone has experienced the same thing?? i have run this model in several computers with enough RAM, memory, 64-bit, etc. but still, i have the same problem....

 

i'm using the 2011 version - moldflow insight. (user for ~5years and i have never seen this problem before)

 

i also thought it was a convergency tolerance problem and changed the convergence to 0.5 with no success. i will try and change the geometric influence from ideal to automatic to see if it improves...

 

any help greatly appreciated....i'm attaching my model for your info...

Message 9 of 29
petr_AL
in reply to: rohitr

Hi,

 

we have 3D study, 5mil. tetras, Cool+Fill+Pack+Warp.

Time for Cooling - about 7days. Job manager had show during this week no progress (100%), but at log file

is possible to see increasing number of iteration - see image.

CPU was working on average 10%. Memory about 80% full.

Time for Fill+Pack+Warp - about 2days.

 

Computer: HP Z800, 2x Xeon 2.67MHz (12 cores), 24GB RAM

Message 10 of 29
yannick.moret
in reply to: petr_AL

If analysis is taking so long, that is moist likely you don't have enough ram on your machine.

I bet you can see constant access on disk and that the cool process is taking only a few % of the CPU.

in the image you attached the part where it shows the required disk space is missing. That is the key.

 

I pasted below my answer

Cool need to write some matrix to solve the thermal solution.

If there is enough space in RAM, it will put it all in RAM, and analysis time will be fast.

If you don't have enoug RAM, it will put it on your hard disk, and then the time will be long as the CPU will have to get the data from it.

It is easy to check, if you look in task manager and see cool is not using 1 core full time, that mean it is runningon disk and will be pretty slow.

 

Look at start of screen output in cool :

Total Mbytes required for Cool analysis = 3628.13

Total Mbytes available for Cool analysis = 420143.38

Now beginning the task: Input part model

Current time is: Thu Aug 11 15:41:52 2011

 

So for example here I need 3.6 gig of memory.

 

Think at closing synergy to save some RAM also while your analysis is running, if you are a bit short on your machine.

Message 11 of 29

@Hugo

See previous answer, check that coo;l is running using 1 full CPU. If not, you are just lacking RAM....

If the analysis ran fast till it seems to stop, then best would be to contact support.

Message 12 of 29

@tobi_cologne

Flow anlaysis is another story (my ram comment apply only for cool).

10 hours can be noraml depending of the size of the model, if you have fiber or not, if you have very long cycle time, if you are close to short shot condition which will lead to convergence difficulties...

Difficult to diagnose without the model.

 

Did you try the latest scnadium lab release with parallelization for midplane and dual domain ?

That would be reallyhelpful if your model has a big number of elements.

Message 13 of 29

Hi Yannick,

 

I'd also think swap file trashing is the culprit, but please read my post above (here I copy it for you convenience):

 

I have a very fast computer with 16GB RAM, and yet the second phase of Cool (FEM) 3D anaysis does take enormous emount of time to complete.

 

Even though the entire sequence is parallelized, and I check the option for using max number of threads - and all 8 threads are reported to be in use - only  the first phase of the Cool (FEM) 3D snalysis can tax my fasy i7 extreme CPU up to 100% for all 8 threads. The second phase - transient within cycle - only uses some 15-18% - and this is not due to low memory and swap file trashing!

 

Just why is that, and what's the bottleneck?

Message 14 of 29

If you have doubt, contact support.

I ran big analysis on cool transient, and saw good performance on it.

It is difficult to do magic reply without model etc...

Message 15 of 29

I did, and just got the confirmation that

 

 

"A Specialist will be assigned to work on your case and will be sending you a status update shortly"

 

Thanks, Autodesk!

 

Piotr

Message 16 of 29

Hello Yannik,

 

thanks for your reply. In the meantime I contacted support and first of all they told me to install service pack 2 because of the bug with getting several times the message "analysis complete". It took just one week to get this update installed 😉 (unfornately this is normal in my company...).

 

We ran my analysis meanwhile on another machine with Xeon 4 x 3,2 Ghz and 12 gb ram (same machine like mine) on AMI2011 and there the analysis (3D, 4,35 mil. elements, cool+fill+pack+warp) took about 3,5 days. But I think this is normal with that high number of elements and too less ram.

 

I'm trying to get an upgrade from 12 gb to 24 or even 36 gb, but at this stage I'll just have to find out a reasonable number of elements (referred to big parts) for getting good results, but without waiting 3 days. 🙂

 

I'll also try out new features for faster analysis.

 

 

Best regards,

 

tobi_cologne 

Message 17 of 29

Any conclusion on this issue..

======================

 

The Time taken for cooling analysis is more than 3 days for 14 lac elements of 3D tetra with BEM method.

i.e mean it is not cool (FEM), running with mold boundry.

 

If RAM size is small, how to speed up the results without compromising tolerance ?

 

Did any one got satisfying answer on the same?

Madhukeshwar Talwar

FORD MOTORS PRIVATE LIMITED, Chennai
mail: madhukeshwart@gmail.com
09600060862
======================================
Please use . Accept as Solution and Give Kudos as appropriate to further enhance these forums. Thank you! .....
Message 18 of 29
teopado89
in reply to: madhukeshwart

Good morning everyone, 

 

i have the same problem:

 

cool+fill+pack+warp analysis too long.

 

If i see on logs right now i have 3 analysis simultaneously:

1)

Total Mbytes required for Cool analysis = 14390.43
Total Mbytes of RAM available for analysis = 7213.94
Total Mbytes of disk space available for analysis = 100075.59
Cool will solve analysis on disk.

 

2)

Total Mbytes required for Cool analysis = 20066.76
Total Mbytes of RAM available for analysis = 16218.60
Total Mbytes of disk space available for analysis = 124546.46
Cool will solve analysis on disk.

 

3)

another...

 

The question is:

if i abort 2 analysis to free space on ram, will the last analysis use the full ram and speed up? of it's useless now because it started on disk and it will work on disk?

Thanks

Message 19 of 29
Idea_Mel
in reply to: teopado89

Hello teopado89,
I think your ram space will free aborting two analyses, so the third job should take advantage.
Message 20 of 29
mppkumar
in reply to: teopado89

Hi

 

Hope you are having more than or equal to 32GB of RAM in your computer

 

In that case, if you abort two analyses it helps to speed up the pending analysis as you have freed RAM usage; But cool analysis doesn't use Parallel solution

 

Whearas COOL (FEM) uses parallel solution (means we can utilize max. number of processors)... this really speeds up your computation time

 

if you dont need transient results you can run only perform steady state analysis (perform analysis with Averaged within cycle option)

 

Link for Cool (FEM) analysis setup: http://help.autodesk.com/view/MFIA/2015/ENU/?guid=GUID-EE4884E1-917A-4AF9-887A-79A13CE3398C

 

for example if we have 10,000,000 tetra elements cool analysis (with simple circuits) may take around 1Hr; same in Cool (FEM) will be completed in less than 30mins.

 

 

 

 

 

 

Thanks
M P Pradeep Kumar


If my views / comments acceptable, provide "Kudos" as appreciation, if answers your query, please click the "Accept as Solution"

Can't find what you're looking for? Ask the community or share your knowledge.

Post to forums  

Autodesk Design & Make Report